RII Track-2 FEC: Building Field-Based Ecophysiological Genome-to-Phenome Prediction

非技术描述：人们普遍认为农作物产量的增长不能满足日益增长的人口的需要。该项目汇集了来自堪萨斯州立大学、俄克拉荷马州立大学和兰斯顿大学（一所历史悠久的黑人大学）的研究人员，以开发一种新的方法来模拟和预测小麦的重要作物生产性状。当前作物性状预测的最大挑战之一是，它落在植物生理学、生物工程、遗传学、计算生物学、数学、统计学和计算机科学之间的一个人口稀少的边缘地带。因此，为了弥补这一差距，将产生结合使用无人机和机器人的观测数据和遗传学数据的数学模型。这些新模型有望简化农民的作物建模，并将有助于农场管理，并且可以很容易地应用于其他作物和其他环境。许多额外的好处也会随之而来。首先，这些数学模型之间的共性意味着结果将很容易转移到许多其他作物上。此外，以这种方式将遗传和观测数据结合起来预测作物性状的好处将有助于农田作物管理，加强粮食安全。针对这些学科的本科生、研究生和教师的教育项目将扩大具有全球竞争力的劳动力队伍。关键企业伙伴的参与也将直接和通过为项目受训者创造市场加速研究向私营部门的转移。最后，购买或制造的传感/测量设备将增强合作伙伴进行广泛的相关数据密集型研究的能力。技术描述：人们普遍认为，到2050年，农作物产量无法达到人类避免重大粮食安全中断所需的产量翻番目标。农民需要基因信息分析来预测管理方案的结果，他们可以在自己独特的田间环境中选择和应用这些管理方案。该项目汇集了来自堪萨斯州立大学、俄克拉荷马州立大学和兰斯顿大学（一所历史悠久的黑人大学）的研究人员，并提出了新的基于小麦生理的作物模型（CMs）。这些CMs将与最先进的现场监测技术与基因组数据相结合，从而重新平衡直接监测与间接模型计算。数据将包括：(1)提取形态特征、冠层温度和光拦截的航空图像。(2)机器人采集2-30 cm（水平/垂直）、3天时间分辨率的多元土壤剖面数据。(3)在64个地点、日期和年份组合中选择的双单倍体系基因表达数据将有助于模型的建立。(4)在堪萨斯州和俄克拉何马州的育种计划中，增加基因型小麦品系的数量也将有助于CM和数量遗传学的整合。如此庞大的数据集通常会给像CMs这样复杂的模型带来计算挑战。与现有的CMs相比，新模型将有效地结合微分方程求解器、最大熵和贝叶斯方法以及高性能计算。结果将是能够在新环境中预测新基因型特征的方法，而不是用于构建模型。许多额外的好处也会随之而来。首先，CMs之间的共性将意味着结果很容易转移到许多其他作物上。此外，提高基因组到表型预测的准确性将有助于农田作物管理，加强粮食安全。面向这些学科的本科生、研究生和教师的教育项目将创造并扩大具有全球竞争力的劳动力队伍。让主要的企业伙伴参与也将直接加速研究转移，并为项目学员创造一个市场。最后，购买或制造的传感/测量设备将增强合作伙伴进行广泛相关数据密集型研究的能力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Nontechnical Description:It is widely agreed that agricultural crop production is not growing to meet the needs of the increasing human population. This project brings together researchers from the Kansas State University, Oklahoma State University, and Langston University (a Historically Black University), to develop a new way to model and predict important crop production traits in wheat. One of the greatest challenges of current crop trait prediction is that it falls in an underpopulated borderland between plant physiology, biological engineering, genetics, computational biology, mathematics, statistics, and computer science. Therefore, to bridge this gap, mathematical models will be produced that combine both observational data using Unmanned Aerial Vehicles and robots, and genetics data. These new models are expected to simplify crop modeling for farmers, and will aid in farm management, and can easily be applied to other crops and in other environments. Many additional benefits will also accrue. First, commonalities between these mathematical models will mean that results will readily transfer to many other crops. Moreover, the benefits of combining genetic and observational data in this way to predict crop traits will aid on-farm crop management, enhancing food security. Educational programs for undergraduates, graduate students, and faculty in these disciplines will enlarge a globally competitive workforce. Involvement of key corporate partners will also speed research transfer to the private sector both directly and by creating a market for project trainees. Finally, the sensing/measuring devices to be bought or built will enhance the ability of partners to conduct a wide range of related, data-intensive research. Technical Description:It is widely agreed that agricultural crop production is not on track to meet the production doubling needed by 2050 for humanity to avoid major food security disruption. Farmers need genetically-informed analytics to predict the outcomes of management options amongst which they may choose and apply in their unique field environments. This project brings together researchers from the sity of Kansas State University, Oklahoma State University, and Langston University (a Historically Black University), and presents new genetically- and physiologically-informed proof-of-concept wheat physiologically-based crop models (CMs). These CMs will link to state-of-the-art field monitoring technologies with genomic data, thus rebalancing direct monitoring vs. indirect model calculation. The data will include: (1) airborne imagery to extract morphological features, canopy temperatures, and light interception. (2) Multivariate soil profile data will be collected by robots at 2-30 cm (horizontal/vertical) and three-day temporal resolution. (3) Gene expression data on selected double haploid lines over 64 combinations of locations, dates, and years will aid in model building. (4) CM and quantitative genetics integration will also be aided by expanding the number of genotyped wheat lines within the Kansas and Oklahoma breeding programs. Such large data sets ordinarily pose computational challenges for models as complex as CMs. In contrast to extant CMs, the new models will efficiently combine differential equation solvers, maximum entropy and Bayesian methods, and high-performance computing. The results will be methods able to predict the traits of novel genotypes in novel environments not used to construct the models. Many additional benefits will also accrue. First, commonalities between CMs will mean that results will readily transfer to many other crops. Moreover, increased genome to phenome prediction accuracy will aid on-farm crop management, enhancing food security. Educational programs for undergraduates, graduate students, and faculty in these disciplines will create and enlarge a globally competitive workforce. Involving key corporate partners will also speed research transfer directly and by creating a market for project trainees. Finally, the sensing/measuring devices to be bought or built will enhance partner ability to conduct a wide range of related, data-intensive research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.